Uhrzeit: 19.45Uhr

 

HT-Tjecken 12800573.1  - 28.04.2009 um 14:15

There is some precious news for you on MyHT space now. One very wet cat is now dry (or at least not in our knees anymore), and another pretty wet cat is also dry. I hope you like these changes (well, fixes really).

The self confidence changes will probably also lead to some better understanding of the game, and explain some of things you sometimes blame random for. The tricky part from our point of view is perhaps that it will look like this change has made attacks much better (as the ratings will be better now that the whole SC effect is included), while it actually has made it a bit harder to score on average.

(and if you say this is the worst implementation ever, I'll slap you around with a really large trout ;))

 

 

HT-Daniel 12800573.8  - 28.04.2009 um 14:22

are the changes in effect starting now? will we be able to see our confidence and the new ratings at our friendly matches today?

 

Yes the new matchengine is active for training matches.

 

 

HT-Tjecken 12800573.10  - 28.04.2009 um 14:24

are the changes in effect starting now? will we be able to see our confidence and the new ratings at our friendly matches today?

 

Yep (but only friendlies this season as written in the MyHT).

 

 

HT-Tjecken 12800573.15  - 28.04.2009 um 14:29

can you give us some details concerning the effect of SC up to now?
numbers would be best, but knowing they won't come, a rough estimate?

 

One can say the the biggest change is on the edges. The bonus for really high SC will be significantly smaller than the current bonus, but the "penalty" for low SC will also be significantly smaller. On "normal" confidence levels, there is of course also a change but far from as big as for extreme levels.

 

 

HT-Tjecken 12800573.35  - 28.04.2009 um 14:46

Will there from now on be some intuitively understandable meaning for the situation of exactly equal attack and defense ratings?

 

Do you mean it's intuitive today, or not?

Anyway, this change will also make equal attack vs defense more understandable when it comes to scoring probability - and also more correct. It helps a lot to say the least to see how good your attack rating really is with all factors included. In the past couldn't really tell how good your attack really was (as confidence also had a hidden impact).

 

 

HT-Tjecken 12800573.37  - 28.04.2009 um 14:50

is it possible to display your own SC in the match reports?

 

We don't do that for team spirit (which affect the midfield in a similar way). It's probably possible though.

 

 

HT-Tjecken 12800573.38  - 28.04.2009 um 14:51

So if my team has higher confidence it's harder to score but if they have low confidence they have a better chance of scoring?

 

Compared to the current situation, what you say is correct.

However, in general it's naturally easier to score the better confidence you have.

 

 

HT-Tjecken 12800573.70  - 28.04.2009 um 16:51

1.) About the statement on average effects (higher ratings, lower scoring probability): Are those global averages over the whole HT?

2.) If scoring probabilities will on average drop slightly, should we expect scoring probability in higher divisions to drop dramatically because they are the teams that have the highest confidence and have thus had the biggest hidden boost?

 

1.) Yes, they're averages over the whole HT. However, we've focused on active teams. Bots are not that interesting in that sense.

2.) Scoring probability will naturally change the most for teams who have very high confidence (and very low confidence, but that's not as common). I would not say it's a dramatical change, that would be a bit overdramatic. But it is a significative change, and the biggest drop (to use a term which perhaps doesn't fit that well) in that sense will be in cases where your attack is weaker than your opponent's defense.

 

 

HT-Tjecken 12800573.148  - 29.04.2009 um 10:06

My second concern, which is actually reflected in the first one, is if this "new feature" will actually have some straight up things or facts. And more clearly, if the random factor is eliminated or reduced.

 

Lets puts things straight. Some people have the impression that the random factor is too big, but that doesn't mean the random factor actually is too big. As a result, it may not help to reduce random - it may still be percieved the other way round. It's therefore more important to focus on your perception of random, in a way explain it.

This change is in line with the above, it's about your perception. As you now can see your true attack ratings without any hidden factor it's easier to make logical assumptions, which will most likely remove some of the "random feeling" you may have. This doesn't mean there will not be any random left, naturally there will be occasions where a rather weak attack score against a much stronger defense (and the other way round) - but in general attack ratings will make more sense than before.

 

Ist der Zufallsfaktor betreffend der Chancenverwertung ganz abgeschafft oder reduziert wurden?

-> Er wurde lediglich „realistischer“ (reduziert) gemacht als zuvor.

 

HT-Tjecken 12800573.149  - 29.04.2009 um 10:10

I think you got it wrong. Bad influence on higher levels means that the power of self-confidence is reduced, but still with a high confidence it's much easier to score than with a lower one.

 

Exactly. High confidence is of course still much better than low confidence, high confidence is just not as good as it's been in the past and low confidence is now a bit better than it's been in the past.

 

 

HT-Tjecken 12800573.150  - 29.04.2009 um 10:11

What I would like to know is if confidence will only affect visible ratings from now on and won´t have an invisible effect.

 

Yes, no hidden effect at all (in friendlies this season, for all matches next season).

 

Also hat das Selbstvertrauen keinen unsichtbaren Effekt mehr auf die Ratings?

-> Genau, es gibt keinen unsichtbaren Effekt mehr auf die Ratings.

 

 

HT-Tjecken 12800573.152  - 29.04.2009 um 10:32

Did confidence have hidden effect also on the probability of chance being reported?

 

Yes, naturally as it was a part of the (true) total attack it also affected the probability of a chance being reported. So, yes this change may lead to fewer chances being reported on average, but there shouldn't be any huge differencies.

 

-> Es kann dazu kommen, dass über weniger Chancen berichtet wird, aber der Unterschied sollte eher klein sein im Gegensatz zu bisher.

 

 

HT-Tjecken 12800573.153  - 29.04.2009 um 10:35

Will we soon get to see the effect of confusion in all the sector ratings, too, or does that one still remain completely off your radar? ;-)

 

The effect of confusion is also included in these more accurate ratings. I guess that's mean it's off our radar now, but in a good way. ;)

 

 

HT-Tjecken 12800573.154  - 29.04.2009 um 10:36

Could all LA's clarify this in their MyHT's? (That it's compared to the past?) Because it's already noticeable how many misunderstandings this is leading too

 

Yep, I'm on my way to add it and tell the LAs.

 

 

HT-Tjecken 12800573.163  - 29.04.2009 um 10:47

The effect of confusion is also included in these more accurate ratings.

----------------------------

This is new. Care to explain?

 

Is it that much to explain really? Confusion affects your ratings, and therefore that effect shold also be visible when it happens. When we now made the whole ratings better we also included this effect.

 

-> Verwirrung ist sichtbar in den Ratings.

 

 

HT-Tjecken 12800573.169  - 29.04.2009 um 11:08

Confusion affects midfield only or not? Because from the ratings it looks like it affects midfield only, but from the reports it looks like it affects other sectors as well.

 

Lets answer this one again, as I've been misinformed (or rather, a darn old code made misinformation possible).

Confusion affects all ratings, but for some fantastic reason it's only been included in the midfield ratings. We will fix this too, making the effect visible for all sectors.

 

Durch einen Bug wurde die Verwirrung nur im Mittelfeld angezeigt, sie betraf aber alle Ratings.

 

 

HT-Tjecken 12800573.219  - 29.04.2009 um 14:29

Here's my match from today: (221557486)

Please note two missed chances from the left and one in the middle for my opponent.

I don't like what I'm seeing there. Are you sure this whole thing works like it should?

 

As I think you know, there's always a possibility that your attack misses a chance (no matter how superior it is compared to the opponent's defense). We've naturally tested this and according to the tests it works as it should. We will naturally continue to analyze the results (I mean, what we're now doing in friendlies is kind of the last step of this test), so why not include this match too.

 

 

HT-Tjecken 12800573.255  - 29.04.2009 um 15:29

Let me sum up factors that, as of yesterday :-), modify team "strength" but unfortunately not team ratings :
a) team self-confidence
b) confusion
c) Special Event "team overconfidence"
d) Special Event "lack of team experience"

a) was fixed for this week's friendly games : nice.
b) will be fixed (cf. your previous message).

What about c) and d) ?

Regards,

-------------------------------

I thought c) is already seen in midfield ratings?

-------------------------------

Sure it is but it is not seen in other ratings although I think it has an impact on them too.

 

Overconfidence is included and only affects midfield. It has no impact on other sectors.

(yes, we've doublechecked that)

 

 

HT-Tjecken 12800573.263  - 29.04.2009 um 15:42

We aren't talking about one chance -- hell, my match could have been "one in a million" (the usual explanation) -- we are talking about three missed chances with attacks that should basically have had almost 100% conversion.

 

Again, there is (and has always been) a slim chance that an attacker misses a chance - even if the attack is a magntude better than the defense. Simply put, there is no 100% conversion guarantee and has never been.

 

 

HT-Tjecken 12800573.268  - 29.04.2009 um 15:50

FYI: "Even more accurate ratings" will be live soon, probably tonight already.

Even more accurate ratings will make ratings accurate (the very meaning of that word), ie include everything which actually affects different sectors.

 

 

HT-Tjecken 12800573.272  - 29.04.2009 um 16:03

Everyone understands that the engine does not have strictly 100% probabilities anywhere, with the possible exception of non-existent ratings. Referring to that completely and utterly misses the point, and only gives the impression that you either refuse to understand what people are talking about, or intentionally want to avoid saying too much.

 

I'm not so sure about your first statement, even in this thread I actually get the impression that people think it's so close that it's actually 100% (more or less). It may naturally be a result of rhetorical choices, but I don't want anyone (who may just read this thread and not participate in it) to get the impression that it should be 100% conversion, got it?

If that's not the case and you're indeed correct in your assumptions of what people understand, well then it's all fine.

 

 

HT-Tjecken 12800573.274  - 29.04.2009 um 16:07

what does that mean? what will be the difference between the ratings that we've seen yesterday at the friendly matches and those that we'll see next week?

 

Not that much, but if you suffer from confusion it will show for all sectors and not midfield only. Pulling back will also show.

 

Wie wird der Unterschied sein zwischen den Ratings von dieser Woche und der nächsten Woche?

-> Nicht groß, außer es gibt Verwirrung oder eine Mannschaft zieht sich zurück nach einer Führung.

 

 

HT-Tjecken 12800573.280 - 29.04.2009 um 16:12

what does that mean? what will be the difference between the ratings that we've seen yesterday at the friendly matches and those that we'll see next week?

--------------------

Not that much, but if you suffer from confusion it will show for all sectors and not midfield only. Pulling back will also show.

--------------------

and Pulling back what affects btw? :)

 

The event pretty much explains itself actually.

 

 

HT-Tjecken 12800573.287 - 29.04.2009 um 16:25

Does it?

 

Actually yes. It's written pretty clearly in the match report what happens, it says something happening with attack and something else with defense. I also think this was a question in our first Ask a HT project, and it's probably wikified since a long time.

 

 

HT-Daniel 12800573.334 - 29.04.2009 um 18:20

Actually I hope it works better than that. It would make very much sense to have a measurement exactly before and exactly after some events (substitution, red card, confusion, etc...).

The difference wouldn't be huge, but at least it would be the most accurate solution.

 

There are measurements before and after subs, red cards, injuries, confusion etc

 

Ja es gibt Messungen der Mannschaftsleistung vor und nach Platzverweisen, Verletzungen, Verwirrungen usw.

 

 

HT-Daniel 12800573.337 - 29.04.2009 um 18:26

The new engine is now up and running. Everyting should now be included in the ratings.
I saw a questing in this thread that someone wondered why it suddenly could take just a few hours to fix a problem that has existed for so many years, the reason for this is that I have been working with this solution for a while and to add these new featues was quite easy when all the ground work was done.

Update 1 – 30.04.2009 - 16.10Uhr

 

HT-Daniel 12800573.358 - 29.04.2009 um 21:29

So this is where you take the data points for the average ratings now. Correct me if I am wrong, but that will be essentially the same method as we have had so far, only with more data points.

The result will be less variation (good), but the same bias as we have had (disappointing). The bias I am referring to is the way the "average event time" in a half match is strongly dependent on the number of events in that half. So, on average we will still have lower average ratings in matches with many events than in matches with few events (due to players being on average more tired at the time things happen).

Would it be that hard to take the measurements at exact 5 minute intervals instead?

 

I dont only measure when an event happen but I use those exact points also.
This new average calculation should give good results.

 

-> Wir benutzen nicht nur Ereignisse, sondern auch feste Punkte zum messen.

 

 

HT-Daniel 12800573.359 - 29.04.2009 um 21:31

Just to confirm:
The new changes apply to YA match engine as well, right? We don't have (visible) confidence, but we do have confusion as well as performance changing over the match due to substitutions.

 

I will add these changes to YA match engine when next season starts.

 

-> Ich werde diese Veränderungen zur neuen Saison der Jugendengine hinzufügen.

 

 

HT-Tjecken 12800573.361 - 29.04.2009 um 21:35

Why not present team ratings using a small (and simple!) graphical representation:

 

Hattrick is a game proudly based on text. We like telling a story in words rather than in graphs for example. That's why we use denominations for example.

Now, I'm not saying graphs is a bad idea as such (on the contrary, but it shouldn't be the "main" representation). We introduced bars for skills in the new design for example, and we also have ideas to give you more feedback on things happening in the game as well as the performance in the game. And together with that graphs would fit better. It does however require a few things more from the match engine, things the engine doesn't have currently.

 

 

HT-Tjecken 12800573.365 - 29.04.2009 um 22:01

The new changes apply to YA match engine as well, right?

 

They don't apply right now (they only apply for senior friendlies now), but they will be next season of course.

Edit: Beaten by Daniel. What the heck is happening in this world??? :)

 

 

HT-Daniel 12800573.366 - 29.04.2009 um 22:04

But if you calculate the ratings every 5 minutes like you said, the only thing you need to implement is a good interface which could present a graphical view of the ratings evolution during the match. You have nothing new to calculate, all is allready implemented, except the graphical view.

I know that certainly means a lot of work, only design work, but I think it could be one of the best improvement you could made for hattrick.

Well, I hope you won't forget this idea :)

 

We dont store all those average calculations in the database today but that might happen in the future.

 

 

HT-Daniel 12800573.384 - 30.04.2009 um 07:25

What kind of ratings changes are you guys seeing so far?

I've been on decent/strong confidence all season, and when I play with 2 TDFs and 1 normal forward I've been around formidable(very high) to outstanding(low). Today, my first friendly after the changes I played like this--plus with 2 offensive IMs.

My confidence is still strong--yet I only managed solid(high).

@Any HT who cares to answer:

That's a loss of over 20% central attack. Was being at decent/strong confidence REALLY inflating my ratings that much?

Edit: I think I may have found my answer. I pulled back after 13 minutes. If I understand correctly, my attack ratings reflect this pulling back. And since the rating displayed is the match average, pulling back early will really lower said rating.

 

You are correct, the reson for the lower ratings are that you pulled back.

 

 

HT-Tjecken 12800573.395 - 30.04.2009 um 09:58

However, in general it's naturally easier to score the better confidence you have.

-----------------------

because the (now fully visible) rating is higher, or because the scoring proability on higher confidence levels is higher?

lets make an example in the new match engine:

team A has a 50% attack vs defense rating and highest confidence level
team B has the same 50% attack vs defence rating but the lowest confidence level.

team A has a greater chance to score than team B despite both teams have the same attack/defense-rating? correct?

 

As both team A and B have the same rating, their chance to score is exactly the same.

Again, the whole confidence effect is now included in the ratings. If you got a high confidence it will lead to better ratings, which in turn increases your chance to score. Shortly put, your confidence affects your attack ratings. Your ratings affects your probability to score. Got it?

               Übersichtliches Archiv der HT-Kommentare

Uhrzeit: 15.15Uhr

Derzeit gibt es einen Bug, der es nicht ermöglicht GB-Einträge zu kommentieren. Dieser tritt schon seit gestern auf, und wurde bereits den GMs gemeldet.

Uhrzeit: 11.22Uhr

Seit einigen Tagen werden bei Nicht-Supportern, vermutlich durch die Werbebanner, die Virenscanner aktiv und melden Viren oder ähnliches. Die HTs haben mittlerweile die Banner abgeschaltet, um dem Problem auf die Schliche zu kommen. Dennoch ist es sehr wichtig, sollte man eine Virenmeldung oder ähnliches in den letzten Tagen auf HT erhalten haben, den PC mal zu durchleuchten.

GM-Visiburn dazu:
12774143.8
Wir haben momentan verschiedene Meldungen über Script-Probleme und malware. Betroffen sind ausschließlich Nicht-Supporter und wir meinen, es hängt tatsächlich mit den Bannern zusammen. Die HTs prüfen das inzwischen, die Banner sollten momentan abgeschaltet sein (was ich leider nicht nachprüfen kann, da ich ja Supporter bin).

Falls bei Euch eine derartige Meldung auftritt, empfehle ich einen Scan Eures PCs (oder was auch immer) mit Adware (z.B. AdAware, etc.). Nett wäre auch, wenn es sich um bösartige Funde handelt, eine Meldung über das Kontaktformular an die GM (/help/contact.aspx).
------------------------
So weit ich aber schon mitbekommen habe, sind das eher seriöse Banner, die Probleme verursachen.

Uhrzeit: 14.45Uhr

 

März 2009:

Einmalige Besucher: 6.601
Angesehene Seiten: 11.872

TOP 3 der meistgelesenen Artikel:
1.) Update für den HO auf 1.424 mit 651 Zugriffen
2.) Vorstellung der Trikot- und Dressdesigner Föderation mit 497 Zugriffen
3.) Update für den HO auf 1.423 mit 385 Zugriffen

Jeweiliger Anteil der Besucher:
- Suchmaschinen: 59%
- Hattrick, Foren usw: 20%
- Direkte Herkunft: 22%
(Die Zahlen sind gerundet.)

Ich danke allen Besuchern und Freunden des Blogs, und hoffe auch auf weiterhin viele Leser in den kommenden Monaten.

Uhrzeit: 10.40Uhr

http://downloads.mozdev.org/foxtrick

Bitte achte auf eventuelle neuere Versionen, da regelmäßig Neue erscheinen.

http://downloads.mozdev.org/foxtrick/foxtrick_0.4.6.2.xpi (<- Direktdownload)


== New features ==
* dropdown image to customlinks
* headerfix for standard layout
* tracker update
* cssfix to get more space for forum tread list in simple skin
* copymatchid added to more matchicons (eg headtohead pages, NT games, youtharchieve, played matchlineup)
* link to team position probability calculator on league pages
* module to add HTlive button to ongoing matches

== Bugfixes ==
* several small fixes related to ht design changes (shortcuts, match preview links, guestbookalltidflags, teampopups at matchpreviewpage)
* interference of playerdetail links and playerslink fixed
* several small fixes (singleheadercheck, copypostid, tempopups with headerfix)
* partly fixing typed message text got lost when going back and forward (meassage is still lost if title edit field visible)
* copy playerad completely in english if ht locale is set to english regardless of foxtrick locale setting
* fixed some multiple tickeralert showing

== known bugs ==
The following bugs are not yet fixed (most need more detailed bug reports. if you encounter one of the following, please open a ticket at http://code.google.com/p/foxtrick/issues/list):

* ColoredYouthFaces may cause firefox to crash on some systems (not fixed, disable module)
* some converted currencies not placed proper (not all fixed)
* Sometimes elements are shown twice (not fixed)
* SubOptions in preferences not shown on first activation of a module (not fixed)
* css-skins may not work well with css-fixes (no solution yet)
* single header line breaking forum design (not a bug, use options to prevent this)
* teampopups flicker for word-wrapped teamnames (not fixed)